4-bit Ripple Carry Adder of Two-Phase clocked Adiabatic Static CMOS Logic: a comparison with static CMOS Nazrul Anuar Yasuhiro Takahashi Toshikazu Sekine Graduate School of Engineering Gifu University, 1-1 Yanagido, Gifu-shi, Gifu Japan 501-1193 Email: n3814101@edu.gifu-u.ac.jp Faculty of Engineering Gifu University, Japan 501-1193 Email: yasut@gifu-u.ac.jp Faculty of Engineering Gifu University, Japan 501-1193 Email: sekine@gifu-u.ac.jp Abstract—The paper presents a new quasi energy recovery logic family that uses two complementary split-level sinusoidal power supply clock for digital low power applications such as sensors. The proposed two-phase adiabatic static CMOS logic circuit (2PASCL) is using the principle of energy recovery and adiabatic switching. It has switching activity that is lower than dynamic logic and can be directly derived from static CMOS circuits. We have designed and simulated a 4-bit ripple carry adder (RCA) using 2PASCL logic gates implemented using 0.18 µm CMOS technology. Driving pulse with the height equal to Vdd is supplied to the gates. From the simulation result, it shows that 2PASCL can save an average of 71.3% of energy dissipation compared with static CMOS logic at transition frequency of 10 to 100MHz. I. I NTRODUCTION With the widely spread of mobile, hand-held and wireless electronics devices, the demands for the innovations of lowpower VLSI arise. For most of the digital circuits today, CMOS logic scheme has been the technology of choice for implementing low-power systems. As the clock and logic speeds increase to meet the new performance requirements, the energy requirement of CMOS circuits are becoming a major concern in the design of above devices. Power dissipation in conventional CMOS primarily occurs during device switching. When a CMOS which consists of the pull-up (pMOS) and pull down (nMOS) networks connected to a load capacitance CL is set into a logical “1” logic level, an energy of Eapplied = C L Vdd 2 is applied to the load [1]. Energy stored is half of the energy supplied, therefore the total dissipation as heat during charging and discharging, when the logic level is “0”, is the same as Etotal = C L Vdd 2 . Energy(dissipation in the channel resistance R is given as ) 2 L Ediss = RC C V L dd . For adiabatic charging, when ∆T , ∆T which means the time for the driving voltage to change from 0 V to Vdd is long, in theory, the energy dissipation is nearly zero. Adiabatic logic circuits also utilize AC power supplies to recycle the energy used to charge node capacitances in the circuit. As it dissipates less energy than the fundamental limit of static CMOS, adiabatic circuits are promising candidates for low-power circuits in the frequency range in which signals are digitally processed. In recent years, studies on adiabatic computing have been utilized for low-power systems and several adiabatic logic families have been proposed [1]– [15]. However, several weaknesses such as complex circuits, multiple power-clock supplies, nonadiabatic transitions that may compromise the energy savings and the logic gates are not well suited for CMOS implementation are seen. In this paper, we propose a Two-Phase Adiabatic Static CMOS Logic (2PASCL) circuit. It can be directly converted from static CMOS circuits without drastically increasing the circuit complexity and transistor overheads. This circuit is emphasized on the recycle of the charges as two diodes are placed for the discharging. The functional of ripple carry adder based on 2PASCL is evaluated and energy dissipation is compared with CMOS at transition frequency of 10 to 100 MHz. II. T WO P HASE A DIABATIC S TATIC CMOS L OGIC Figure 1 shows a circuit diagram and waveforms illustrating the operation of the 2PASCL inverter. It resembles the static CMOS logic inverter but operates in a nearly adiabatic fashion. The first difference of 2PASCL compared to static CMOS logic gate is the two diodes, one from the output node to the power clock supply and another one is placed next to the nMOS logic to another power clock. The pMOS and nMOS diodes are used to recycle the charge from the output node. The other difference is that split-level sinusoidal power clock supplies, ϕ and ϕ are used to replace the Vdd and the Vss . From the simulation, we found that split-level sinusoidal gives a lower energy dissipation compared to trapezoidal power clock supply even if we set the Trise and Tf all of the trapezoidal waveforms to a maximum values. By using two split-level sinusoidal waveforms which each peak-to-peak is 0.9 V, we can reduced the voltage difference, thus reducing the charging and discharging activities. Sinusoidal waveforms can also be generated with higher energy efficiency than trapezoidal waveforms [15]. The operation of 2PASCL is as follows. Let us consider the inverter logic circuit demonstrated in Fig. 1. The circuit φ operation phase is divided into evaluation and hold. In evaluation phase, when the output node Y is LOW and pMOS tree is turned ON, as ϕ swings up and ϕ swings down, CL is charged through pMOS transistor resulting the HIGH state at the output. When Y is LOW and nMOS is ON, no transition occurs. The same result gained when the output node is HIGH and pMOS is ON. When Y node is HIGH and nMOS is ON, discharging via nMOS and D2 resulting the output voltage decreased to Vt value before entering hold mode where the logical state is “0” [12]. For the evaluation mode where ϕ swings up and ϕ swings down, the detailed operations are summarized by Table I. ON transistor M1 M2 M1 M2 III. S IMULATION RESULTS AND DISCUSSION A. Condition of simulation The paper starts by examining the functional and energy dissipation of a simple logic gate, an inverter of 2PASCL as shown in Fig. 1. We use SPICE simulation with a 0.18 µm -1.8 V standard CMOS process. The W/L of nMOS and pMOS logic gates used is 0.6 µm/0.18 µm. A capacitive load CL , of 0.01 pF is placed at output node Y. Bulks of pMOS and nMOS are connected to ϕ and ϕ accordingly. The power supply clock frequencies are set to be 4 times higher than the transition frequency after considering energy dissipation results. Using split-level sinusoidal power clock driving voltage of 0.9 V peak-to-peak each, the result at 100 MHz transition frequency of 2PASCL inverter is shown in Fig. 1. The input signals are CMOS-compatible rectangular pulses. The top graph shows the split-level sinusoidal supply clock driving voltage. The second graph demonstrates the input signal and the third graph shows the output waveform of a correctly functioning inverter which was simulated with the SPICE. Its energy dissipation is calculated by integrating the voltage and current product value as follows: ) ∫ Ts (∑ n E= (Vpi × Ipi ) dt, (1) 0 i=1 φ Y M2 CL φ D2 φ 1.9V 1.4V 0.9V 0.4V -0.1V 1.9V 1.4V 0.9V 0.4V -0.1V Next state of Y HIGH LOW (no transition) HIGH (no transition) LOW At hold mode where ϕ swings down and ϕ swings up, due to the diodes, the state of Y when preliminary state is LOW remains unchanged. When the preliminary state of the output node is HIGH, it will change to Vt , the threshold voltage of the diode. At this point, discharging via D1 occurs. From the operation of 2PASCL, less dynamic switchings are seen as circuit nodes are not necessarily charging and discharging every clock cycle which reduces the node switching activities significantly. The lower the switching activity reduces energy dissipation. φ X TABLE I O PERATION SUMMARY AT EVALUATION MODE Prelim. state Y LOW LOW HIGH HIGH D1 M1 V(-phi) φ V(phi) V(x) V(y) 1.6V 1.1V 0.6V 0ns Fig. 1. 10ns 20ns 30ns 40ns 50ns 2PASCL inverter circuit and its waveforms. where Ts is the period of the primary input signal, Vp is the power supply voltage, Ip is the power supply current and n is the number of power supply [12]. The energy in joule is then converted to watt by multiplying it with the input frequency. TABLE II C IRCUIT DATA FOR 2PASCL AND STATIC CMOS Driving power voltage Split level ϕ and ϕ Frequency (input: driving voltage) nMOS, pMOS (incl. diodes) COMPARISON 0–1.8V 0–0.9V, 0.9–1.8V 1:4 W/L : 0.6µm/0.18µm s a b cin cout Fig. 2. Full adder. Then, the simulation of 2PASCL based 1-bit full adder (FA) is carried out. As shown in Fig. 2, FA consists of exclusiveOR and NAND logic gates. Each exclusive-OR and NAND of 2PASCL is as demonstrated in Fig. 3 and Fig. 4 respectively. To verify the applicabilities of the proposed structure and 2PASCL property, a 4-bit ripple carry adder (RCA) has been designed and simulated. To enhance the performance evaluation, we simulate the frequency characteristics of power consumption for RCA of 2PASCL and the result is compared with CMOS. The transition frequencies simulated are from 10 to 100 MHz. The energy dissipation value is taken at the same time period for both circuits. φ a Y b B. Results and discussion From the simulations, we have confirmed the functional of the 4-bit ripple carry adder 2PASCL logic circuits. The results are shown in Fig. 5. From the results, 2PASCL with split level sinusoidal clocking voltage gives a significant lower energy dissipation compared with conventional static CMOS for ripple carry adder. It also demonstrates a significant lower energy dissipation when transition frequency is simulated from 10 to 100 MHz which is shown in Fig. 6. φ Fig. 4. Schematic for NAND logic of 2PASCL. Y a φ φ b Fig. 3. Schematic for exclusive-OR logic of 2PASCL. 2.0V 1.0V 0.0V 2.0V 1.0V 0.0V 2.0V 1.0V 0.0V 2.0V 1.0V 0.0V 2.0V 1.0V 0.0V 2.0V 1.0V 0.0V 0ns V(-phi) V(phi) V(a) V(b) V(cin) V(s) V(cout) 180ns 360ns 540ns 720ns 900ns Fig. 5. Output waveforms for ripple carry adder of 2PASCL from the simulation result. IV. C ONCLUSION R EFERENCES This paper has described a simulation of Two-Phase clocked Adiabatic Static CMOS Logic (2PASCL). By implementing the adiabatic charging and energy recovery theory, 4-bit ripple carry adder (RCA) of 2PASCL is compared to RCA of static CMOS. Compared with static CMOS circuits, 2PASCL consumes much less power. For instance, at input frequency of 10 to 100 MHz, RCA of 2PASCL dissipates only 28.7% of the energy of static CMOS logic in average. From the results of other logic gates such as inverter chain [16], NAND, exclusive-OR and FA of 2PASCL which have been simulated, this logic scheme can be a viable candidate for ultra-low energy computing. [1] J. Marjonen, and M. Aberg, “A single clocked adiabatic static logic - a proposal for digital low power applications,” J. VLSI Signal Processing, vol. 27, no. 2–3, pp. 253–268, Feb. 2001. [2] V.I. Starosel’skii, “Adiabatic logic circuits: A review,” Russian Microelectronics, vol.31, no. 1, pp.37–58, 2002 [3] W.C. Athas, L.J. Svensson, J.G. Koller, N. Tzartzanis and E.Y.-C. Chou, “Low-power digital systems based on adiabatic-switching principles”, IEEE Trans. VLSI Syst., vol.2, no.4, pp.398–407, 1994. [4] K.A. Valiev and V.I. Starosel ’skii, “A model and properties of a thermodynamically reversible logic gate,” Mikroelektronika, vol.29, no.2, pp.83–98, 2000 [5] S.G. Younis and T.F. Knight, “Asymptotically zero energy split-level charge recovery logic,” Proc. 1994 Int. Workshop on Low Power Design, pp.177–182, 1994. [6] V.I. Starosel ’skii, “Reversible logic,” Mikroelektronika, vol.28, no.3, pp.213–222, 1999. ENERGY DISSIPATION COMPARISON OF RIPPLE CARRY ADDER: 2PASCL VERSUS CMOS Energy dissipation, [uW] 160 CMOS - RCA 80 2PASCL - RCA 40 20 10 20 Fig. 6. 40 Transition frequency, [MHz] Energy dissipation comparison with different transition frequency. [7] J.S. Denker, “A review of adiabatic computing,” Symp. on Low Power Electronics, San Diego, 1994. [8] A.G. Dickinson and J.S. Denker, “Adiabatic dynamic logic,” Proc. IEEE CICC, pp.282–285, 1994. [9] A.G. Dickinson and J.S. Denker, “Adiabatic dynamic logic,” IEEE J. Solid-State Circuits, vol.30, no.3, pp.311–315, 1995. [10] Y. Moon and D.-K. Jeong, “An efficient charge recovery logic circuit,” IEEE J. Solid-State Circuits, vol.31, no.4, pp.514–522, 1996. [11] K. Takahashi and M. Mizunuma, “Adiabatic dynamic CMOS logic circuit,” IEICE Trans. Electron. (Japanese Edition), vol.J81-CII, no.10, pp.810–817, Oct. 1998 (Electronics and Communications in Japan Part II (English Translation), vol.83, no.5 , pp.50–58, April 2000). [12] Y. Takahashi, Y. Fukuta, T. Sekine and M. Yokoyama, “2PADCL : Two phase drive adiabatic dynamic CMOS logic”, Proc. IEEE APCCAS, pp. 1486-1489 ,Dec 2006 [13] Alan Kramer, John S. Denker, Stephen C. Avery, Alex G. Dickinson and Thomas R. Wik, “Adiabatic computing with the 2N-2N2D logic family,” 1994 Symposium on VLSl Circuits Digest of Technical Papers, pp.25–26, 1994 [14] C.L. Seitz, A.H. Frey, S. Mattisson, S.D. Rabin, D.A. Speck and J.L.A. Van de Snepscheut, “Hot-clock nMOS”, 1985 Chapel Hill Conference on Very Large Scale Integration, Fuchs, H., Ed., Rockville, Md.: Computer Science, pp.1–17, 1985 [15] Y. Ye and K. Roy, “QSERL: Quasi-static energy recovery logic,” IEEE J. Solid-States Circuits, vol.36, no.2, pp.239–248, Feb. 2001. [16] Nazrul Anuar, Y. Takahashi and T. Sekine, “Two phase clocked adiabatic static logic circuit: a proposal for digital low power applications,” Proc. IEICE Gen. Conf., pp.102, Mar. 2009. 80